
import random
import pandas as pd
import numpy as np
import seaborn as sns
import matplotlib.pyplot as plt

%matplotlib inline


df = pd.read_csv("../data/raw/Employee-Attrition.csv")
df.head()


df.shape

(1470, 35)


def generate_meta_signals(df):
    meta_df = df.isna().sum().reset_index()
    meta_df['% of Total Values'] = meta_df[0]/len(df)
    meta_df = pd.concat([meta_df, pd.DataFrame(df.dtypes, columns=['dtypes']).reset_index(drop=True)], axis=1)
    d = df.describe().T.reset_index()
    meta_df = meta_df.merge(d, on=['index'], how='left')
    vcs = [pd.DataFrame(df[x].value_counts(normalize=True)).T for x in 
    list(df.columns)]
    vcs= [pd.DataFrame((x.idxmax(axis=1), x.max(axis=1))).T.reset_index() for x in vcs if len(list(x.columns)) > 0]
    meta_df = meta_df.merge(pd.concat(vcs), on=['index'], how='left') 
    meta_df.columns = ['Index', 'Missing Values', '% of Total Values', 'dtypes', 'count', 'mean', 'std', 'min', '25%', '50%', '75%', 'max', 'Most Frequently Value','% of MF Values']
    return meta_df


def bad_flag(row):
    if row['std'] == 0:
        return True
#     elif (row['25%'] == row['50%']) and (row['50%'] == row['75%']):
#         return True
    elif row['% of Total Values'] > .15:
        return True
    elif row['% of MF Values'] > .80:
        return True
    elif pd.isnull(row['% of MF Values']):
        return True
    else:
        return False


meta_df = generate_meta_signals(df)
    
meta_df['is Trash var'] = meta_df.apply(lambda r: bad_flag(r), axis=1)
meta_df.head(35)


df.drop(['EmployeeCount', 'Over18', 'StandardHours'], axis=1, inplace=True)


from sklearn.preprocessing import LabelEncoder
map_list = []
le_name_mapping = dict()
le = LabelEncoder()
df["Attrition"] = le.fit_transform(df["Attrition"])
le_name_mapping["Attrition"] = dict(zip(le.classes_, le.transform(le.classes_)))


import plotly.express as px


fig = px.pie(df, "Attrition", title="Attrition distribution", hole=.3)
fig.show()


attr_corr = df[(df.select_dtypes(include=np.number).columns.difference(['Attrition']))].corrwith(df["Attrition"]).sort_values(ascending=False)
df_attr_corr = pd.DataFrame(attr_corr)


fig = px.imshow(df_attr_corr, color_continuous_scale='Viridis')
fig.update_layout(
    height=800
)
fig.show()


df_corr = df[(df.select_dtypes(include=np.number).columns.difference(['Attrition']))].corr()
text = df_corr.values.tolist()
fig = px.imshow(df_corr, color_continuous_scale='Viridis', aspect="auto")
fig.update_xaxes(side="top")
fig.show()


fig = px.histogram(df, x="TotalWorkingYears", color="Attrition")
fig.show()


df["TotalWorkingYears"].corr(df["Attrition"])

-0.17106324613622667


fig = px.histogram(df, x="Age", color="Attrition")
fig.show()


df["Age"].corr(df["Attrition"])

-0.1592050068657797


fig = px.histogram(df, x="Gender", color="Attrition", barmode="group")
fig.show()


fig = px.histogram(df, x='BusinessTravel', color='Attrition', barmode="group")
fig.show()


fig = px.histogram(df, x='OverTime', color='Attrition', barmode="group")
fig.show()


pd.crosstab(df['OverTime'], df['Gender'])


fig = px.histogram(df, x="MonthlyIncome", color="Attrition")
fig.show()


df[df.columns.difference(["Attrition"])].head(2).to_csv("../data/request.csv", index=[0])


from sklearn.model_selection import train_test_split
from sklearn.feature_extraction import DictVectorizer

from sklearn.metrics import roc_auc_score
from sklearn.preprocessing import MinMaxScaler


# categorical_columns = [col for col in df.columns if df[col].dtype == 'object']
# print(categorical_columns)
# for cat_col in categorical_columns:
#     le = LabelEncoder()
#     df[cat_col] = le.fit_transform(df[cat_col])
#     le_name_mapping[cat_col] = dict(zip(le.classes_, le.transform(le.classes_)))


# le_name_mapping


df_train, df_test = train_test_split(df, test_size=.2, random_state=1)


y_train, y_test = df_train['Attrition'], df_test['Attrition']
for df_ in [df_train, df_test]:
    del df_['Attrition'] 
y = y_train.append(y_test, ignore_index = True)


train_dict = df_train.to_dict(orient='records')
test_dict = df_test.to_dict(orient='records')


# sparse - is a matrix that is comprised of mostly zero values.
dv = DictVectorizer(sparse=False)
dv.fit(train_dict)

DictVectorizer(sparse=False)

DictVectorizer(sparse=False)


X_train = dv.transform(train_dict)
X_test = dv.transform(test_dict)
X = np.concatenate((X_train, X_test))


dv.get_feature_names()

C:\Users\User\Anaconda\lib\site-packages\sklearn\utils\deprecation.py:87: FutureWarning:

Function get_feature_names is deprecated; get_feature_names is deprecated in 1.0 and will be removed in 1.2. Please use get_feature_names_out instead.

['Age',
 'BusinessTravel=Non-Travel',
 'BusinessTravel=Travel_Frequently',
 'BusinessTravel=Travel_Rarely',
 'DailyRate',
 'Department=Human Resources',
 'Department=Research & Development',
 'Department=Sales',
 'DistanceFromHome',
 'Education',
 'EducationField=Human Resources',
 'EducationField=Life Sciences',
 'EducationField=Marketing',
 'EducationField=Medical',
 'EducationField=Other',
 'EducationField=Technical Degree',
 'EmployeeNumber',
 'EnvironmentSatisfaction',
 'Gender=Female',
 'Gender=Male',
 'HourlyRate',
 'JobInvolvement',
 'JobLevel',
 'JobRole=Healthcare Representative',
 'JobRole=Human Resources',
 'JobRole=Laboratory Technician',
 'JobRole=Manager',
 'JobRole=Manufacturing Director',
 'JobRole=Research Director',
 'JobRole=Research Scientist',
 'JobRole=Sales Executive',
 'JobRole=Sales Representative',
 'JobSatisfaction',
 'MaritalStatus=Divorced',
 'MaritalStatus=Married',
 'MaritalStatus=Single',
 'MonthlyIncome',
 'MonthlyRate',
 'NumCompaniesWorked',
 'OverTime=No',
 'OverTime=Yes',
 'PercentSalaryHike',
 'PerformanceRating',
 'RelationshipSatisfaction',
 'StockOptionLevel',
 'TotalWorkingYears',
 'TrainingTimesLastYear',
 'WorkLifeBalance',
 'YearsAtCompany',
 'YearsInCurrentRole',
 'YearsSinceLastPromotion',
 'YearsWithCurrManager']


numerical_vars = list(df[df.columns.difference(["Attrition"])].select_dtypes(include=np.number).columns)


aucs = dict()
for var in numerical_vars:
    auc_ = roc_auc_score(y_train, df_train[var])
    if auc_ < 0.5:
        auc_ = roc_auc_score(y_train, -df_train[var])
    aucs[var] = auc_


sorted(aucs.items(), key=lambda x: x[1])

[('MonthlyRate', 0.501490505034657),
 ('Education', 0.5064663263533618),
 ('HourlyRate', 0.5100384953743913),
 ('PerformanceRating', 0.510694093453545),
 ('EmployeeNumber', 0.5113496915326987),
 ('RelationshipSatisfaction', 0.514397942430644),
 ('NumCompaniesWorked', 0.5167317595243832),
 ('TrainingTimesLastYear', 0.5218028386836487),
 ('PercentSalaryHike', 0.5256747897323254),
 ('YearsSinceLastPromotion', 0.5261370704291646),
 ('DailyRate', 0.5464466023769633),
 ('JobSatisfaction', 0.5560480323652521),
 ('WorkLifeBalance', 0.56864448092882),
 ('EnvironmentSatisfaction', 0.5697231358881112),
 ('JobInvolvement', 0.5741946509920823),
 ('DistanceFromHome', 0.5742198663028191),
 ('StockOptionLevel', 0.6228742092198383),
 ('Age', 0.6255666440662772),
 ('YearsWithCurrManager', 0.6292396743302534),
 ('JobLevel', 0.6324560272997765),
 ('YearsInCurrentRole', 0.6384432627491413),
 ('TotalWorkingYears', 0.6402727736281469),
 ('YearsAtCompany', 0.6428167183113587),
 ('MonthlyIncome', 0.6477112902954674)]


min_max = MinMaxScaler()
min_max.fit(X_train)
X_train_scaled = min_max.transform(X_train)
X_test_scaled = min_max.transform(X_test)
X_scaled = np.concatenate((X_train_scaled, X_test_scaled))


from sklearn.linear_model import LogisticRegression
from sklearn.svm import SVC
# from sklearn.neighbors import KNeighborsClassifier
from sklearn.tree import DecisionTreeClassifier
from sklearn.ensemble import RandomForestClassifier

from catboost import CatBoostClassifier
from xgboost import XGBClassifier

from sklearn.model_selection import KFold
from sklearn.metrics import f1_score, precision_score, recall_score, precision_recall_curve


log_clf = LogisticRegression()
svc_clf = SVC()
# knn_clf = KNeighborsClassifier()
dt_clf = DecisionTreeClassifier()
rf_clf = RandomForestClassifier()
cat_clf = CatBoostClassifier()
xgb_clf = XGBClassifier()


kfold = KFold(n_splits=5, shuffle=True, random_state=1)
scores = dict()

for clf in [log_clf, svc_clf, dt_clf, rf_clf, cat_clf, xgb_clf]:
    clf_list = []
    for train_idxs, test_idxs in kfold.split(X):
        X_train, X_test = X[train_idxs], X[test_idxs]
        y_train, y_test = y[train_idxs], y[test_idxs]

        clf.fit(X_train, y_train)
        y_pred = clf.predict(X_test)
        clf_list.append(roc_auc_score(y_test, y_pred))
    scores[clf.__class__.__name__] = clf_list


for key in scores.keys():
    print(f'{key}' + ' ' + f'{np.mean(scores[key]):.3f} +- {np.std(scores[key]):.3f}')


kfold = KFold(n_splits=5, shuffle=True, random_state=1)
scores = dict()

for clf in [log_clf, svc_clf, dt_clf, rf_clf, cat_clf, xgb_clf]:
    clf_list = []
    for train_idxs, test_idxs in kfold.split(X_scaled):
        X_train_scaled, X_test_scaled = X_scaled[train_idxs], X_scaled[test_idxs]
        y_train_scaled, y_test_scaled = y[train_idxs], y[test_idxs]

        clf.fit(X_train_scaled, y_train_scaled)
        y_pred = clf.predict(X_test_scaled)
        clf_list.append(roc_auc_score(y_test_scaled, y_pred))
    scores[clf.__class__.__name__] = clf_list


for key in scores.keys():
    print(f'{key}' + ' ' + f'{np.mean(scores[key]):.3f} +- {np.std(scores[key]):.3f}')


from sklearn.model_selection import RandomizedSearchCV


solvers = ['newton-cg', 'lbfgs', 'liblinear', 'sag', 'saga']

penalty = ['none', 'l1', 'l2', 'elasticnet']

c_values = [100, 10, 1.0, 0.1, 0.01]

lr_grid = dict(solver=solvers, penalty=penalty, C=c_values)
print(lr_grid)


lr_random = RandomizedSearchCV(log_clf, param_distributions=lr_grid, scoring='roc_auc', cv=5, n_iter=20,
                               random_state=1, n_jobs=-1, verbose=1)
lr_random.fit(X_train_scaled, y_train_scaled)


y_pred_lr = lr_random.predict_proba(X_test_scaled)[:, 1]


print(lr_random.best_score_)
print(lr_random.best_params_)


# n_neighbors = [int(x) for x in np.arange(1,22,1)]

# metric = ['eucledian', 'manhattan', 'minkowski']

# weights = ['uniform', 'distance']

# knn_grid = dict(n_neighbors=n_neighbors, weights=weights, metric=metric)
# print(knn_grid)


# knn_random = RandomizedSearchCV(knn_clf, param_distributions=knn_grid, scoring='roc_auc', cv=5, n_iter=20,
#                                random_state=1, n_jobs=-1, verbose=1)
# knn_random.fit(X_train_scaled, y_train_scaled)


# y_pred_knn = knn_random.predict_proba(X_test_scaled)[:, 1]


# print(knn_random.best_score_)
# print(knn_random.best_params_)


kernel = ['poly', 'rbf', 'sigmoid']
C = [50, 10, 1.0, 0.1, 0.01]
gamma = ['scale']

svc_grid = dict(kernel=kernel,C=C,gamma=gamma)
print(svc_grid)


svc_random = RandomizedSearchCV(svc_clf, param_distributions=svc_grid, scoring='roc_auc', cv=5, n_iter=20,
                               random_state=1, n_jobs=-1, verbose=1)
svc_random.fit(X_train_scaled, y_train_scaled)


print(svc_random.best_score_)
print(svc_random.best_params_)


#number of trees in forest
n_estimators = [int(x) for x in np.linspace(100,1200,12)]
#max depth of tree
max_depth = [int(x) for x in np.linspace(5,30,6)]
#quality of split
criterion = ['gini','entropy']
#min no. of samples to consider for splitting a internal node
min_samples_split = [2,5,7,10]
#min number of node can be as leaf node
min_samples_leaf = [2,5,8]
# The number of features to consider when looking for the best split:
max_features = ["auto","sqrt"]

random_grid = dict(n_estimators=n_estimators, max_depth=max_depth, criterion=criterion,
                  min_samples_split=min_samples_split,min_samples_leaf=min_samples_leaf,
                   max_features=max_features)
print(random_grid)


rf_random = RandomizedSearchCV(rf_clf, param_distributions=random_grid, scoring='roc_auc', cv=5, n_iter=20,
                               random_state=1, n_jobs=-1, verbose=1)

rf_random.fit(X_train_scaled, y_train_scaled)


y_pred_rf = rf_random.predict_proba(X_test_scaled)[:, 1]


print(rf_random.best_score_)
print(rf_random.best_params_)


n_estimators = [int(x) for x in np.linspace(100, 1000, 10)]

max_depth = [int(x) for x in np.linspace(6, 30, 5)]

learning_rate = [x for x in np.arange(0.001, 0.4, 0.001)]

l2_leaf_reg = [x for x in np.arange(0, 4, 0.1)]

min_data_in_leaf = np.random.randint(20, 200, 5)

cat_grid = dict(n_estimators=n_estimators, max_depth=max_depth, learning_rate=learning_rate,
                l2_leaf_reg=l2_leaf_reg, min_data_in_leaf=min_data_in_leaf)


# cat_random = RandomizedSearchCV(cat_clf, param_distributions=cat_grid, cv=5, random_state=1,
#                                 n_iter=20, scoring="roc_auc", n_jobs=-1, verbose=1)
# cat_random.fit(X_train_scaled,y_train_scaled)


# y_pred_cat = cat_random.predict_proba(X_test_scaled)[:, 1]


# print("score: ", cat_random.best_score_)
# print("best_params: \n", cat_random.best_params_)


n_estimators = [int(x) for x in np.linspace(100, 1000, 10)]

max_depth = [int(x) for x in np.linspace(6, 30, 5)]

learning_rate = [x for x in np.arange(0.001, 0.4, 0.001)]

# how many observations we need to have in a leaf node
min_child_weight = list(range(1, 10))

xg_grid = dict(n_estimators=n_estimators, max_depth=max_depth, learning_rate=learning_rate,
               min_child_weight=min_child_weight)
print(xg_grid)


xgb_random = RandomizedSearchCV(xgb_clf, param_distributions=xg_grid, cv=5, random_state=1,
                                n_iter=20, scoring="roc_auc", n_jobs=-1, verbose=1)
xgb_random.fit(X_train_scaled, y_train_scaled)


y_pred_xgb = xgb_random.predict_proba(X_test_scaled)[:, 1]


print("score: ",xgb_random.best_score_)
print("best_params: \n",xgb_random.best_params_)


thresholds = np.linspace(0, 1, 101)


def confusion_matrix_dataframe(y_val, y_pred):
    scores = []
    for t in thresholds:
        actual_positive = (y_val == 1)
        actual_negative = (y_val == 0)

        predict_positive = (y_pred >= t)
        predict_negative = (y_pred < t)

        tp = (predict_positive & actual_positive).sum()
        tn = (predict_negative & actual_negative).sum()

        fp = (predict_positive & actual_negative).sum()
        fn = (predict_negative & actual_positive).sum()

        tpr = tp/(tp + fn)
        fpr = fp/(fp + tn)
        
        precision = tp / (tp + fp)
        recall = tp / (tp + fn)
        
        f1 = 2 * precision * recall / (precision+recall)
        
        gmean = np.sqrt(tpr * (1-fpr))
        
        auc = (1+tpr-fpr)/2
        
        scores.append((t, tp, fp, fn, tn, tpr, fpr, precision, recall, f1, gmean, auc))

    columns = ['threshold', 'tp', 'fp', 'fn', 'tn', 'tpr', 'fpr', 'precision', 'recall', 'f1', 'gmean', 'auc']
    df_scores = pd.DataFrame(scores, columns=columns)
    
    return df_scores


df_scores = confusion_matrix_dataframe(y_test_scaled, y_pred_lr)
df_scores.head()


import plotly.graph_objects as go


# locate the index of the largest g-mean
ix = np.argmax(df_scores.gmean)
print('Best Threshold=%f, G-Mean=%.3f' % (df_scores.threshold[ix], df_scores.gmean[ix]))


df_scores[df_scores.precision==df_scores.recall]


fig = go.Figure()
fig.add_trace(go.Scatter(x=df_scores.threshold, y=df_scores.precision,
                    mode='lines',
                    name='precision'))
fig.add_trace(go.Scatter(x=df_scores.threshold, y=df_scores.recall,
                    mode='lines',
                    name='recall'))
fig.add_vline(0.26, line_dash="dot")
fig.show()


fig = go.Figure()
fig.add_trace(go.Scatter(x=df_scores.threshold, y=df_scores.f1,
                    mode='lines',
                    name='lines'))
fig.add_vline(0.25, line_dash="dot")
fig.show()


fig = go.Figure()
fig.add_trace(go.Scatter(x=df_scores.fpr, y=df_scores.tpr,
                    mode='lines'))
fig.update_layout(
    title="ROC Curve",
    xaxis_title="False Positive Rate",
    yaxis_title="True Positive Rate",
)
fig.add_vline(0.15, line_dash="dot")
fig.show()


!jupyter nbconvert --to html "EDA, Modelling + Tuning.ipynb"

	Index	dtypes	count	mean	std	min	25%	50%	75%	max	Most Frequently Value	% of MF Values	is Trash var
0	Age	int64	1470.0	36.923810	9.135373	18.0	30.00	36.0	43.00	60.0	35.0	0.053061	False
1	Attrition	object	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	No	0.838776	True
2	BusinessTravel	object	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	Travel_Rarely	0.709524	False
3	DailyRate	int64	1470.0	802.485714	403.509100	102.0	465.00	802.0	1157.00	1499.0	691.0	0.004082	False
4	Department	object	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	Research & Development	0.653741	False
5	DistanceFromHome	int64	1470.0	9.192517	8.106864	1.0	2.00	7.0	14.00	29.0	2.0	0.143537	False
6	Education	int64	1470.0	2.912925	1.024165	1.0	2.00	3.0	4.00	5.0	3.0	0.389116	False
7	EducationField	object	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	Life Sciences	0.412245	False
8	EmployeeCount	int64	1470.0	1.000000	0.000000	1.0	1.00	1.0	1.00	1.0	1.0	1.0	True
9	EmployeeNumber	int64	1470.0	1024.865306	602.024335	1.0	491.25	1020.5	1555.75	2068.0	2048.0	0.00068	False
10	EnvironmentSatisfaction	int64	1470.0	2.721769	1.093082	1.0	2.00	3.0	4.00	4.0	3.0	0.308163	False
11	Gender	object	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	Male	0.6	False
12	HourlyRate	int64	1470.0	65.891156	20.329428	30.0	48.00	66.0	83.75	100.0	66.0	0.019728	False
13	JobInvolvement	int64	1470.0	2.729932	0.711561	1.0	2.00	3.0	3.00	4.0	3.0	0.590476	False
14	JobLevel	int64	1470.0	2.063946	1.106940	1.0	1.00	2.0	3.00	5.0	1.0	0.369388	False
15	JobRole	object	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	Sales Executive	0.221769	False
16	JobSatisfaction	int64	1470.0	2.728571	1.102846	1.0	2.00	3.0	4.00	4.0	4.0	0.312245	False
17	MaritalStatus	object	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	Married	0.457823	False
18	MonthlyIncome	int64	1470.0	6502.931293	4707.956783	1009.0	2911.00	4919.0	8379.00	19999.0	2342.0	0.002721	False
19	MonthlyRate	int64	1470.0	14313.103401	7117.786044	2094.0	8047.00	14235.5	20461.50	26999.0	4223.0	0.002041	False
20	NumCompaniesWorked	int64	1470.0	2.693197	2.498009	0.0	1.00	2.0	4.00	9.0	1.0	0.354422	False
21	Over18	object	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	Y	1.0	True
22	OverTime	object	NaN	NaN	NaN	NaN	NaN	NaN	NaN	NaN	No	0.717007	False
23	PercentSalaryHike	int64	1470.0	15.209524	3.659938	11.0	12.00	14.0	18.00	25.0	11.0	0.142857	False
24	PerformanceRating	int64	1470.0	3.153741	0.360824	3.0	3.00	3.0	3.00	4.0	3.0	0.846259	True
25	RelationshipSatisfaction	int64	1470.0	2.712245	1.081209	1.0	2.00	3.0	4.00	4.0	3.0	0.312245	False
26	StandardHours	int64	1470.0	80.000000	0.000000	80.0	80.00	80.0	80.00	80.0	80.0	1.0	True
27	StockOptionLevel	int64	1470.0	0.793878	0.852077	0.0	0.00	1.0	1.00	3.0	0.0	0.429252	False
28	TotalWorkingYears	int64	1470.0	11.279592	7.780782	0.0	6.00	10.0	15.00	40.0	10.0	0.137415	False
29	TrainingTimesLastYear	int64	1470.0	2.799320	1.289271	0.0	2.00	3.0	3.00	6.0	2.0	0.372109	False
30	WorkLifeBalance	int64	1470.0	2.761224	0.706476	1.0	2.00	3.0	3.00	4.0	3.0	0.607483	False
31	YearsAtCompany	int64	1470.0	7.008163	6.126525	0.0	3.00	5.0	9.00	40.0	5.0	0.133333	False
32	YearsInCurrentRole	int64	1470.0	4.229252	3.623137	0.0	2.00	3.0	7.00	18.0	2.0	0.253061	False
33	YearsSinceLastPromotion	int64	1470.0	2.187755	3.222430	0.0	0.00	1.0	3.00	15.0	0.0	0.395238	False
34	YearsWithCurrManager	int64	1470.0	4.123129	3.568136	0.0	2.00	3.0	7.00	17.0	2.0	0.234014	False

| Import Data¶

| Data Wrangling¶

| Data Validation¶

| EDA¶

Target variable¶

Correlation with target variable¶

Correlation matrix¶

Experience¶

Age¶

Gender¶

Travel¶

Overtime¶

Montly Income¶

Save data for sender¶

| Data Processing for Training¶

Feature Engineering¶

Label Encoder¶

Train-test Split¶

One-hot encoding¶

Feature Importance¶

Feature Selection¶

Feature Scaling¶

| Modelling¶

Performance tuning (hyperparameters)¶

LR¶

KNN¶

SVC¶

RF¶

CatB¶

XGB¶

Create a confusion matrix¶

Find threshold¶

| Make html¶

	Age	Attrition	BusinessTravel	DailyRate	Department	DistanceFromHome	Education	EducationField	EmployeeCount	EmployeeNumber	...	RelationshipSatisfaction	StandardHours	StockOptionLevel	TotalWorkingYears	TrainingTimesLastYear	WorkLifeBalance	YearsAtCompany	YearsInCurrentRole	YearsSinceLastPromotion	YearsWithCurrManager
0	41	Yes	Travel_Rarely	1102	Sales	1	2	Life Sciences	1	1	...	1	80	0	8	0	1	6	4	0	5
1	49	No	Travel_Frequently	279	Research & Development	8	1	Life Sciences	1	2	...	4	80	1	10	3	3	10	7	1	7
2	37	Yes	Travel_Rarely	1373	Research & Development	2	2	Other	1	4	...	2	80	0	7	3	3	0	0	0	0
3	33	No	Travel_Frequently	1392	Research & Development	3	4	Life Sciences	1	5	...	3	80	0	8	3	3	8	7	3	0
4	27	No	Travel_Rarely	591	Research & Development	2	1	Medical	1	7	...	4	80	1	6	3	3	2	2	2	2

Gender	Female	Male
OverTime
No	408	646
Yes	180	236